Third and fifth grade children (average age 8.6 and 10.6 years) and adult participants were tested with printed words of varying length in a new on-line identification task (the luminance increment paradigm, LIP) and a speeded naming task. Effects of general length (length in letters, phonemes and syllables) were shown to decrease systematically with age in both tasks. Third grade children showed substantial effects of word length while the effect practically disappeared in adults. In general, this developmental pattern was also found when separately examining effects of phonological length (with length in letters held constant) and small unit length (with number of syllables held constant), although some differences were observed in performance in the identification and the naming task. The two tasks also showed different developmental patterns, with the greatest gain in performance arising between 3rd and 5th grade in the naming task, and the largest improvement occurring between 5th grade and adults in the identification task. The results suggest that the new luminance increment paradigm can be usefully applied as an on-line measure of printed word perception in beginning readers.